Resolve accuracy issue from fusing prelu #29754
Open
+33
−11
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Min, Byungil <byungil.min@intel.com>
Signed-off-by: Min, Byungil <byungil.min@intel.com>
byungilm
changed the title
| @@ -742,6 +742,12 @@ void prepare_primitive_fusing::fuse_simple_primitives(program &p) { | |||
| // prelu fusion is not implemented in oneDNN3.1 (CVS-108233) | |||
| return; | |||
| } | |||
|
|
|||
| if (activation_func == cldnn::activation_func::relu_negative_slope && | |||
| input.get_output_layout().batch() > 1 && input.is_type<convolution>()) { | |||
There was a problem hiding this comment.
In case of dynamic layout .batch() will lead to exception
Additional check is required
Signed-off-by: Min, Byungil <byungil.min@intel.com>
Signed-off-by: Min, Byungil <byungil.min@intel.com>
Signed-off-by: Min, Byungil <byungil.min@intel.com>
isanghao
reviewed
Mar 27, 2025
| @@ -3786,7 +3803,7 @@ TEST_P(post_ops_optimizations_onednn_eltw_any_sum_eltw_linear, basic) { | |||
| data("out_hi", get_mem(get_single_element_layout(p), 127)), | |||
| data("eltwise_data", get_mem(get_output_layout(p))), | |||
| convolution("conv_prim", input_info("input"), "weights", "bias", p.groups, p.stride, p.dilation, p.pad, p.pad, format::is_grouped(get_weights_layout(p).format)), | |||
| activation("activation", input_info("conv_prim"), activation_func::relu_negative_slope), | |||
| activation("activation", input_info("conv_prim"), activation_func::relu), | |||
There was a problem hiding this comment.
could you explain why you changed the function to relu from relu_negative_slope?
There was a problem hiding this comment.
changed activation becuz some cases are (multi batch conv and prelu) condition. After check the test, it changed value of expected_fused count for onednn.
| convolution_test_params{ CASE_MUL_BATCH_CONV_U8S8_1, 4, 4, 4 }, | ||
| convolution_test_params{ CASE_MUL_BATCH_CONV_U8S8_2, 4, 4, 4 }, | ||
| convolution_test_params{ CASE_MUL_BATCH_CONV_S8S8_1, 4, 4, 4 }, | ||
| convolution_test_params{ CASE_MUL_BATCH_CONV_S8S8_2, 4, 4, 4 }, |
There was a problem hiding this comment.
Is it necessary to add such many test cases? Isn't it enough to add just one?
There was a problem hiding this comment.
Reduced test-cases.
src/plugins/intel_gpu/src/graph/graph_optimizer/prepare_primitive_fusing.cpp
Show resolved
Hide resolved
Signed-off-by: Min, Byungil <byungil.min@intel.com>
isanghao
reviewed
Mar 28, 2025
| #define CASE_MUL_BATCH_CONV_U8S8_1 { 32, 15, 4, 5 }, { 32, 30, 2, 3 }, { 30, 15, 3, 3 }, { 1, 1 }, { 0, 0 }, { 1, 1 }, 1, data_types::u8, format::bfyx, data_types::i8, format::bfyx, data_types::f32, format::bfyx | ||
| #define CASE_MUL_BATCH_CONV_U8S8_2 { 20, 15, 5, 5 }, { 20, 30, 3, 3 }, { 30, 15, 3, 3 }, { 1, 1 }, { 0, 0 }, { 1, 1 }, 1, data_types::u8, format::bs_fs_yx_bsv16_fsv16, data_types::i8, format::os_is_yx_osv16_isv16, data_types::f32, format::bfyx | ||
| #define CASE_MUL_BATCH_CONV_S8S8_1 { 32, 15, 4, 5 }, { 32, 30, 2, 3 }, { 30, 15, 3, 3 }, { 1, 1 }, { 0, 0 }, { 1, 1 }, 1, data_types::i8, format::bfyx, data_types::i8, format::bfyx, data_types::f32, format::bfyx | ||
| #define CASE_MUL_BATCH_CONV_S8S8_2 { 20, 15, 5, 5 }, { 20, 30, 3, 3 }, { 30, 15, 3, 3 }, { 1, 1 }, { 0, 0 }, { 1, 1 }, 1, data_types::i8, format::bs_fs_yx_bsv16_fsv16, data_types::i8, format::os_is_yx_osv16_isv16, data_types::f32, format::bfyx |
There was a problem hiding this comment.
please remove unused macros.
isanghao
approved these changes
Mar 28, 2025
github-merge-queue bot
pushed a commit
that referenced
this pull request
Mar 28, 2025
### Details: - blocked fusing condition conv+prelu ### Tickets: - CVS-163143 --------- Signed-off-by: Min, Byungil <byungil.min@intel.com>
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Details:
Tickets: